基于GWO-LMS-RSSD的旋转机械耦合故障分离及特征强化方法  

Coupling fault separation and feature enhancement method for rotatingmachinery based on GWO-LMS-RSSD

作　　者：许文[1] 施卫华[2] 李红钢[2] 华如南 刘厚林[1] 董亮[1] XU Wen;SHI Weihua;LI Honggang;HUA Runan;LIU Houlin;DONG Liang(National Research Center of Pumps,Jiangsu University,Zhenjiang 212013,China;Wuhan Second Ship Design and Research Institute,Wuhan 430060,China)

机构地区：[1]江苏大学国家水泵及系统工程技术研究中心,江苏镇江212013 [2]武汉第二船舶设计研究所,湖北武汉430060

出　　处：《机电工程》2025年第4期677-685,共9页Journal of Mechanical & Electrical Engineering

基　　金：国家自然科学基金资助项目(52279087,51879122);江苏省自然科学基金资助项目(BK20241801);泰州市重大科技成果转化项目(SCG202205)。

摘　　要：针对旋转机械耦合故障中较弱故障易被较强故障淹没及噪声干扰严重的问题,提出了基于灰狼优化算法(GWO)的自适应滤波最小均方(LMS)算法,结合共振稀疏分解(RSSD)的耦合故障特征分离及强化方法。首先,采用自适应滤波LMS算法对耦合故障信号进行了滤波处理,使故障特征得到了初步强化;然后,根据耦合故障的不同共振属性,利用RSSD算法将故障耦合分解为高共振分量和低共振分量,完成了耦合故障分离;特别地,针对LMS算法中参数依赖人工经验、自适应差等问题,研究了基于灰狼优化算法(GWO)的参数自适应优化方法,设计了以信噪比和均方误差构成的优化目标;最后,对稀疏分解得到的信号进行了包络解调,完成了耦合故障分离及特征强化,同时,利用模拟信号和实验信号对该方法进行了验证分析。研究结果表明:GWO-LMS-RSSD算法能用于有效降低噪声干扰,分离旋转机械耦合故障及强化故障特征。该研究成果可为强噪声干扰下耦合故障的特征分离及强化提供一种新的思路。Aiming at the issue of weaker faults being overshadowed by stronger faults and severe noise interference in rotating machinery coupling faults,a method combining the adaptive least mean square(LMS)filtering algorithm with resonance-based sparse signal decomposition(RSSD)was proposed for feature separation and enhancement of coupling faults.Firstly,the adaptive filtering LMS algorithm was applied to filter coupling fault signals,providing initial enhancement of fault features.Then,based on the different resonance properties of coupling faults,the RSSD algorithm was utilized to decompose the coupled faults into high-resonance components and low-resonance components,achieving fault separation.In particular,to address the issues of parameter reliance on manual experience and poor adaptiveness in the LMS algorithm,a parameter self-adaptive optimization method based on the grey wolf optimizer(GWO)was investigated,with optimization objectives designed using signal-to-noise ratio and mean squared error.Finally,envelope demodulation was performed on the signals obtained from sparse decomposition to achieve fault separation and feature enhancement.Simulated and experimental signals were used to verify and analyze the proposed method.The research results demonstrate that the GWO-LMS-RSSD algorithm can effectively reduce noise interference,separate coupled faults in rotating machinery,and enhance feature extraction,offering a new approach for fault feature separation and enhancement under strong noise interference.

关 键 词：耦合故障诊断 旋转机械 共振稀疏分解 自适应滤波最小均方算法 灰狼优化算法 信噪比 均方误差 

分 类 号：TH133[机械工程—机械制造及自动化]